めちゃくちゃわかりやすいAlphaFoldの資料ですね

Congratulations to @DemisHassabis and John Jumper who received the 2023 Breakthrough Prize in Life Sciences for their work on #AlphaFold , an AI system to predict the 3D structure of proteins. 🧬

Find out more. → dpmd.ai/3JFJgKV

Many congratulations to our CEO Demis Hassabis and colleague John Jumper at Google DeepMind, who were honoured this weekend with a Breakthrough Prize in Life Sciences for their work on AlphaFold. Read more here: bit.ly/3UhPlmO #AI #machinelearning

ScaleFold: Reducing AlphaFold Initial Training Time to 10 Hours
arxiv.org/html/2404.1106…

He's referring to alphafold, which cannot solve protein folding FFS!!!!!!!! This is the problem with many AI doomer types. They sound knowledgeable and like they know what they're talking about, until they discuss something in an area you know about at which point...

Had an amazing time at #HTCC6 in beautiful Dubrovnik! 🏝️
Loved the crystal fishing workshops & late-night microseedings! Great discussions on molecular replacement with #CCP4 . Thrilled to share #AlphaFold ’s impact on solving protein structures. Can't wait for next year!

Our Progres method for fast protein structure searching can now search the AlphaFold database TED domains at Dali-level accuracy in a tenth of a second per query on CPU. (1/5)

biorxiv.org/content/10.110…

New preprint out!🚨We performed a large scale analysis of physicochemical features extracted from over half a million AlphaFold structural models, and using data-driven methods we showed a link between these features and in vivo behaviour of proteins. Find out more below⬇️1/9

Quite detailed and unique dissection of the AlphaFold algorithm!

I love Alphafold! It may just reveal the location of a key factor on a giant complex for the first time, with high confidence score! #alphafold

Matthew Barnett Eliezer Yudkowsky ⏹️ yeah alphafold didn't even solve protein folding, it is very much an approximation and it's not close to being 100% trustworthy. the only reason we use it is that empirically determining protein structure (by cryo EM or crystallography) is very painstaking.

AlphaFold was an incredible achievement, but it was only possible with 'billions of year of PhD's time' to generate the data and knowledge it was built on

Solutions like AlphaFold model direct physicochemical interactions between amino acids to determine the most likely 3D structure of a protein but do not account for the cellular processes, like the action of chaperones, that can affect protein folding in vivo.

the irony of saying something like that right after quote tweeting a tour de force cryo-EM paper

'oh yes, alphafold actually already solved this, we just wanted to do it the hard way for fun'

Unlocking enzyme secrets with OSMES: our structure-based bioinformatics screening leverages AlphaFold models for identifying orphan enzymatic reactions 🔬🧬 #Bioinformatics #PathwayHoles . Just published: nature.com/articles/s4146…

xyx Ruxandra Teslo 🧬 Rohan 🌱 🇧🇷 Presumably it would mean a physics-driven description of folding. Alphafold is a statistical model of which sequences tend to coevolve and therefore which ones are likely associated in space. No actual folding is simulated.

It was fun sharing our work ❄️🐉 with the CSML community earlier this week! Also fun giving a guest lecture with the same title but different content ( #AlphaFold ++) this morning Princeton Chemistry

I'll be returning to MIT next Tues for a lecture on some combination of the above🤩

Predictomes: A classifier-curated database of AlphaFold-modeled protein-protein interactions
Interesting PPI virtual screening work based on AlphaFold-Multimer.
- SPOC (Structure Prediction and Omics informed Classifier) is a random forest-based classifier that accurately

AlphaFold lecture slides are here: amyxlu.github.io/data/alphafold…

Truly believe in AlphaFold as an AlexNet moment for biology, to endure for its empirical validation of the computational ingredients it propounds rather than (just) as a tool. Super fun to do this deep dive 🧬

Love too claim X-risk based on half-understood marketing copy! !

AlphaFold explicitly *cannot* solve protein folding, it creates a predictive model of structure from based on other known shapes with *zero process steps* PROTEIN FOLDING IS STILL AN OPEN PROBLEM